A Novel Algorithmic Approach for Privacy-Preserving Data Synthesis via Randomized Mixing

EasyChair Preprint 15513

Abstract

Data privacy is a critical concern in modern data-sharing ecosystems. This paper introduces a novel algorithm, RMD-Mix (Randomized Mixing for Differential Privacy), designed to enhance privacy preservation in synthetic dataset generation. By leveraging randomized transformations and controlled perturbation mechanisms, RMD-Mix achieves strong privacy guarantees while retaining high utility for downstream tasks. Extensive experiments on real-world datasets demonstrate the efficacy of RMD-Mix in maintaining privacy and usability, outperforming existing differential privacy-based synthesis methods.

Keyphrases: Privacy-Preserving Algorithms, Randomized Mixing, data privacy, dataset synthesis, differential privacy

@booklet{EasyChair:15513,
  author    = {Michael Lornwood and Li Wei and Emily Wilson},
  title     = {A Novel Algorithmic Approach for Privacy-Preserving Data Synthesis via Randomized Mixing},
  howpublished = {EasyChair Preprint 15513},
  year      = {EasyChair, 2024}}